FBetaMeasure metric with one value per key

[eraldoluis]

Fixes #5637 .

Changes proposed in this pull request:

• I have implemented a class called FBetaMeasure2 that returns a dictionary where each key has only one value.
• This implementation also allows the return of the per-class values and the averages (at the same time).

Before submitting

• [X] I've read and followed all steps in the Making a pull request section of the CONTRIBUTING docs.
• [X] I've updated or added any relevant docstrings following the syntax described in the Writing docstrings section of the CONTRIBUTING docs.
• [ ] If this PR fixes a bug, I've added a test that will fail without my fix.
• [X] If this PR adds a new feature, I've added tests that sufficiently cover my new functionality.

After submitting

• [ ] All GitHub Actions jobs for my pull request have passed.
• [ ] codecov/patch reports high test coverage (at least 90%). You can find this under the "Actions" tab of the pull request once the other checks have finished.

[eraldoluis]

Thank you @dirkgr for reviewing this PR. I made the suggested changes.

[dirkgr]

It seems to be failing tests. I put an updated version at https://github.com/allenai/allennlp/pull/5651 that fixes some things, but it looks like the actual test for the new metric is failing.

[eraldoluis]

It seems to be failing tests. I put an updated version at #5651 that fixes some things, but it looks like the actual test for the new metric is failing.

Thank you, @dirkgr , for pointing this issue. There were two tests missing adaptation from the older version. They have now passed in my tests.

I have also incorporated your edits from #5651 .

[dirkgr]

Thank you for sticking with it!

[eraldoluis]

Thank you for your support!